Self-organizing neural network model of motion processing in the visual cortex during smooth pursuit

Area of Science:

• Computational neuroscience
• Systems neuroscience
• Visual processing

Background:

• Cortical motion processing is crucial for visually guided behaviors like pursuit eye movements.
• Understanding how the brain integrates visual and motor information remains a challenge.

Purpose of the Study:

• To develop a physiologically based neural network model of cortical motion processing during pursuit eye movements.
• To investigate the mechanisms underlying the transition from retinal to real-world reference frames in the middle-superior-temporal (MST) area.

Main Methods:

• Constructed a three-layer neural network simulating primary visual cortex (V1), middle-temporal area (MT), and MST.
• Employed an unsupervised training process for MST unit connections.
• Analyzed model connectivity to understand network functions.

Main Results:

• The model successfully simulated information processing across V1, MT, and MST areas.
• MST units demonstrated integration of visual and eye-movement information.
• Developed MST units showed a transition from retinal to real-world reference frames.

Conclusions:

• The neural network model provides insights into cortical mechanisms for motion processing during pursuit.
• The findings highlight the role of MST in reference frame transformation.
• Model connectivity analysis elucidates functional principles of the network.